Autonomous investing system uses multiple agents and self-improves safely
Governed Self-Evolution for Autonomous Investing: A Multi-Agent LLM System with an Invariant Core
Artificial IntelligenceComputation and LanguageMultiagent Systems
Summary
Making investment decisions automatically from live web data is tricky because web content can be untrustworthy and actions can't be undone. The authors built Steward, a system where nine specialized AI agents analyze different financial information together and manage risks carefully. The system remembers past decisions, grades its own trades, and can improve itself without weakening important safeguards. They tested Steward live for five months and found it limited losses better than a similar single-agent setup and was able to learn from its mistakes.
What this means in practice
- •For quantitative investment teams: Use a multi-agent system with automated review and controlled self-improvement to manage autonomous trading with reduced risk and better response to market data.
- •For financial risk managers: Integrate a risk auditing agent and invariant core rules within a complex AI system to ensure trading actions respect predefined limits and reduce drawdown.
Authors
Yebo Feng, Jingyang Dai, Liangrui Pan, Nelson Lau, Jiahua Xu, Yang Liu
Abstract
LLM agents that act on live Web data must combine heterogeneous evidence, remember what they have done and improve from outcomes without weakening their own safeguards. We study this problem in autonomous investing, where every action is irreversible and based on untrusted Web content. We present Steward, a multi-agent LLM system that draws the boundary of self-evolution in its type system. First, nine role-specialised agents decide jointly: four analysts read news feeds, SEC filings, macroeconomic series and market data and emit typed reports, from which a Portfolio Manager sets target positions that a protected Risk Manager audits. Second, a seven-layer shared memory, assembled by code from the system's own record, gives each decision its position context, the market narrative, calibration and the agent's past decisions. Third, the system improves itself by review: code grades every trade and forecast nightly and writes the grades into memory, and each quarter a reflector proposes prompt learnings that a deterministic gate applies reversibly and replay of recorded inputs verifies. All three operate around an invariant core of order arithmetic, risk rules and discipline-agent prompts that no self-edit can reach. In a five-month live deployment (102 trading days, 1,632 LLM calls), the system contained a 24.6% sector drawdown to a 0.3% loss; in replay of its recorded inputs, a single model given the same raw data traded more and lost 1.35 points against the market over five days where the joint design showed no significant loss, and no injected instruction was followed; and a 2015-2026 backtest shows the core alone lowering the maximum drawdown under all 50 random entry policies. The reflector traced the dominant failure to its root cause, and the gate applied a learning that the agent adopted on trigger evenings and on no control evening. Code: https://github.com/yebof/quant-agent